日本語はこちら

TSUMUGI (Trait-driven Surveillance for Mutation-based Gene module Identification) is a tool that uses phenotype data from knockout mice provided by the International Mouse Phenotyping Consortium (IMPC).
It identifies and visualizes gene modules based on phenotypic similarity.

TSUMUGI is available as a web-tool that anyone can access from here👇️

🔗 https://larc-tsukuba.github.io/tsumugi/

The name TSUMUGI is derived from the Japanese word 紡ぎ, symbolizing the concept of twisting various elements (genes) together to create a unified entity (phenotype).

📖 How to Use TSUMUGI

💬 Top Page

TSUMUGI accepts the following three types of input:

1. Phenotype

When you input a phenotype of interest, TSUMUGI searches for a group of genes whose knockout (KO) mice not only exhibit that phenotype, but also share similar overall phenotype profiles.
Phenotype names are based on the Mammalian Phenotype Ontology (MPO).

You can view the list of phenotypes currently searchable with TSUMUGI here:
👉 Phenotype List

2. Gene

When you specify a particular gene, TSUMUGI searches for other genes whose KO mice share similar phenotype profiles with the KO mouse of the specified gene.
Gene symbols follow the official nomenclature from MGI.

You can view the list of gene symbols currently searchable with TSUMUGI here:
👉 Gene Symbol List

3. Gene List

You can input multiple genes (one per line).
TSUMUGI calculates phenotypic similarity between genes within the list only.

Note

Unlike a single gene search, the Gene List input compares phenotypes within the listed genes only.

Caution

If no similar phenotype pairs are found, the message
No similar phenotypes were found among the entered genes. will appear.
If more than 200 gene pairs are detected, an alert will show
Too many genes submitted. Please limit the number to 200 or fewer. to prevent browser overload.

📥 Download Data

You can download raw data of phenotype similarities between gene pairs (CSV format, gzip-compressed).

The file includes:

• Gene pair (Gene1, Gene2)
• Phenotypic similarity score (Jaccard Similarity)
• Number of shared phenotypes
• List of shared phenotypes

Caution

The file size is approximately 100MB. Downloading may take some time.

🌐 Network Visualization

After input is submitted, the page automatically transitions to a network view.

Only gene pairs that share three or more abnormal phenotypes are visualized.

Nodes

• Each node represents a single gene.
• Clicking a node displays the list of abnormal phenotypes observed in its KO mouse.
• Nodes can be repositioned freely by dragging.

Edges

• Clicking an edge reveals detailed information about the shared phenotypes between the two connected genes.

Control Panel

The left-hand control panel allows you to adjust the appearance and content of the network.

Change Network Layout

You can choose from the following layout options:

• Cose (default)
• Circle
• Random
• Grid
• Concentric

Filter by Phenotype Similarity

The Phenotypes similarity slider lets you filter gene pairs based on their phenotypic similarity (Jaccard index).
The similarity scores are scaled to a range of 1 to 10, allowing filtering in 10 steps.

Filter by Phenotype Severity

The Phenotypes severity slider allows you to filter nodes based on phenotype severity in KO mice.
Higher severity scores indicate stronger phenotypic effects.
This score is also scaled from 1 to 10 for 10-step filtering.

Note

The severity slider is not available if the phenotype data from IMPC is binary (e.g., abnormal embryo development) or if the input consists of a single gene.

Adjust Network Appearance

The following visual properties can be adjusted:

• Font size
• Edge thickness
• Node repulsion (distance between nodes; applicable to Cose layout only)

Search Gene

You can search for specific gene names within the network.

Specify Genotype

You can restrict results to phenotypes observed in KO mice with a specific genotype:

• Homo: Homozygous
• Hetero: Heterozygous
• Hemi: Hemizygous

Specify Sex

You can extract phenotypes specific to sex:

• Female
• Male

Export

You can export the current network as PNG and CSV files.
The CSV file includes cluster (connected component) information and a list of phenotypes observed for each gene's KO mouse.

---

🔍 Method for Identifying Phenotypically Similar Gene Groups

Data Source

The dataset used is IMPC Release-22.1 (2024-12-11), specifically the statistical-results-ALL.csv.gz file.
Detailed information about the data fields can be found here:
👉 Data fields

Preprocessing

Gene–phenotype pairs were extracted where the KO mouse exhibited a phenotype with a p-value ≤ 0.0001
(p_value, female_ko_effect_p_value, or male_ko_effect_p_value).

• Genotype-specific phenotypes are annotated with homo, hetero, or hemi.
• Sex-specific phenotypes are annotated with female or male.

Phenotype Similarity Calculation

Phenotypic similarity is quantified using the Jaccard index,
which measures the proportion of shared phenotypes between two gene knockouts as a value between 0 and 1.

Jaccard(A, B) = |A ∩ B| / |A ∪ B|

For example, suppose KO mice of gene A and gene B exhibit the following abnormal phenotypes:

A: {abnormal embryo development, abnormal heart morphology, abnormal kidney morphology}
B: {abnormal embryo development, abnormal heart morphology, abnormal lung morphology}

In this case, there are 2 shared phenotypes and 4 unique phenotypes in total.
The Jaccard index is calculated as follows:

Jaccard(A, B) = 2 / 4 = 0.5

✉️ Contact

If you have any questions or requests, feel free to contact us using one of the methods below (inquiries in Japanese are welcome):

• Google Form
  👉 Contact Form

• If you have a GitHub account
  👉 GitHub Issue